|
|
Boost Testing : |
From: Rene Rivera (grafik.list_at_[hidden])
Date: 2005-12-07 16:23:54
Beman Dawes wrote:
> "Martin Slater" <mslater_at_[hidden]> wrote in message
> news:438D9BEC.4040105_at_hellinc.net...
>
>>Eric Niebler wrote:
>>
>>>For the [mslater] mingw tests, xpressive is failing due to what looks to
>>>be a configuration problem. See http://tinyurl.com/87dg9:
>>>
>>>c:/mingw/bin/../lib/gcc-lib/mingw32/3.2.3/../../../libmingw32.a(main.o)(.text+0x97):main.c:
>>>undefined reference to `WinMain_at_16'
>>>
>>>FWIW, xpressive's tests pass for the Bronek mingw toolset.
> The g++ command line is missing -L"c:/cygwin/lib/mingw", or whatever your
> appropriate path is.
>
> The obvious workaround is to add an environmental variable:
>
> set MINGW_STDLIB_DIRECTORY=c:\cygwin\lib\mingw
>
> That almost works, but also adds a spurious -L"" to the g++ command line. If
> you remove the -L"" by hand, the g++ command then works OK.
>
> We need help from a Boost.Build expert:
>
> * Why does MINGW_STDLIB_DIRECTORY have to be set by hand?
It's one of the standard MinGW config options for the toolset. But it
would be preferable to set MINGW_ROOT_DIRECTORY, which will the cause
the various GCC_*_DIRECTORY vars.
> Is the problem
> that cygwin installs the mingw libraries in lib\mingw rather than lib?
Don't know... Is the real installed toolset MinGW, or is it CYGWIN?
Although from the above command (I'm getting pagenot found errors for
the regression results currently) it looks like it's a regular MinGW.
> Do we
> need a separate cygwin-mingw toolset?
No, it should be possible to use the mingw toolset directly for that, as
it adds the -mno-cygwin option.
> * Where is the spurious -L"" coming from and how do we get rid of it?
Not sure, since I can't see the original page :-(
-- -- Grafik - Don't Assume Anything -- Redshift Software, Inc. - http://redshift-software.com -- rrivera/acm.org - grafik/redshift-software.com -- 102708583/icq - grafikrobot/aim - Grafik/jabber.org